Orientation constraint for link 'tip' is probably incorrect warning

Question

Hi, I was planning a pose for a 5-dof manipulator tip with moveit kdl solver. The position of the pose is reached, but the orientation cannot be reached. And the following warning is shown every time:

[ INFO] [1488121271.999191010, 373.663000000]: Planning request received for MoveGroup action. Forwarding to planning pipeline.
[ WARN] [1488121271.999485048, 373.663000000]: Orientation constraint for link 'tip' is probably incorrect: 0.000000, 0.000000, 1.000000, 0.500000. Assuming identity instead.
[ WARN] [1488121271.999603296, 373.663000000]: Orientation constraint for link 'tip' is probably incorrect: 0.000000, 0.000000, 1.000000, 0.500000. Assuming identity instead.
[ INFO] [1488121271.999718045, 373.663000000]: No optimization objective specified, defaulting to PathLengthOptimizationObjective
[ INFO] [1488121271.999857147, 373.663000000]: LBKPIECE1: Starting planning with 1 states already in datastructure
[ INFO] [1488121272.444189499, 374.101000000]: LBKPIECE1: Created 38 (20 start + 18 goal) states in 37 cells (19 start (19 on boundary) + 18 goal (18 on boundary))
[ INFO] [1488121272.444250950, 374.101000000]: Solution found in 0.444438 seconds
[ INFO] [1488121272.467475751, 374.124000000]: SimpleSetup: Path simplification took 0.000019 seconds and changed from 23 to 2 states
[ INFO] [1488121272.500154361, 374.156000000]: Combined planning and execution request received for MoveGroup action. Forwarding to planning and execution pipeline.
[ WARN] [1488121272.500245398, 374.156000000]: Orientation constraint for link 'tip' is probably incorrect: 0.000000, 0.000000, 1.000000, 0.500000. Assuming identity instead.
[ INFO] [1488121272.500362095, 374.156000000]: Planning attempt 1 of at most 1
[ WARN] [1488121272.500788898, 374.157000000]: Orientation constraint for link 'tip' is probably incorrect: 0.000000, 0.000000, 1.000000, 0.500000. Assuming identity instead.
[ WARN] [1488121272.500939599, 374.157000000]: Orientation constraint for link 'tip' is probably incorrect: 0.000000, 0.000000, 1.000000, 0.500000. Assuming identity instead.
[ INFO] [1488121272.501062772, 374.157000000]: No optimization objective specified, defaulting to PathLengthOptimizationObjective
[ INFO] [1488121272.501254698, 374.157000000]: LBKPIECE1: Starting planning with 1 states already in datastructure
[ INFO] [1488121273.689715240, 375.327000000]: LBKPIECE1: Created 75 (37 start + 38 goal) states in 74 cells (37 start (37 on boundary) + 37 goal (37 on boundary))
[ INFO] [1488121273.690050718, 375.327000000]: Solution found in 1.188844 seconds
[ INFO] [1488121273.700934373, 375.345000000]: SimpleSetup: Path simplification took 0.000024 seconds and changed from 66 to 2 states
[ WARN] [1488121273.702429142, 375.345000000]: Dropping first 1 trajectory point(s) out of 10, as they occur before the current time.
First valid point will be reached in 0.060s.

Does anybody have any idea what is wrong, please?

---

Originally posted by yangtfu on ROS Answers with karma: 11 on 2017-02-26

Post score: 1

Answer 1

The cause of the warning lies in the Quaternion/Orientation values moveit received as goal. Quaternions that represent orientations have to be normalized to 1. This is not true for the quaternion you passed in: (0.0,0.0,1.0,0.5). As a result MoveIt ignores it and internally replaces it by the identity (0.0,0.0,0.0,1.0).

Side note: You might want to set RRTConnectkConfigDefault as your PlannerId. LBKPIECE often yields pretty weird/infeasible paths with MoveIt at the moment.

---

Originally posted by v4hn with karma: 2950 on 2017-02-26

This answer was ACCEPTED on the original site

Post score: 2

---

Original comments

Comment by yangtfu on 2017-02-26:
Do you mean the quaternion vector including w should be normalized? I thought the axis vector of xyz should be a normalized vector which represents the rotation axis. What's the point of normalizing a quaternion vector.

Comment by v4hn on 2017-02-26:
Quaternions are not an Axis/Angle representation. They are a completely separate form of represention of Orientations. If you want to learn more about this, please look at the appropriate Math literature. For many people I know it is enough though to think of them as a normalized 4-vector.

Comment by yangtfu on 2017-02-26:
Yeah, I misunderstood the concept, thank you!

Comment by Beshari on 2021-01-28:
Had the same error, now I understand, thanks for the RRTConnectkConfigDefault suggestion.